The Relative Weight of Temporal Envelope Cues in Different Frequency Regions for Mandarin Sentence Recognition
The Relative Weight of Temporal Envelope Cues in Different Frequency Regions for Mandarin Sentence Recognition
复制标题
普通话句子识别中不同频率区域时间包络线索的相对权重
DOI:
10.1155/2017/7416727
复制
发表时间:
2017-01-01
影响因子:
3.1
通讯作者:
Yin, Shankai
中科院分区:
文献类型:
--
作者:
Guo, Yang;Sun, Yuanyuan;Yin, Shankai
Acoustic temporal envelope (E) cues containing speech information are distributed across the frequency spectrum. To investigate the relative weight of E cues in different frequency regions for Mandarin sentence recognition, E information was extracted from 30 contiguous bands across the range of 80–7,562 Hz using Hilbert decomposition and then allocated to five frequency regions. Recognition scores were obtained with acoustic E cues from 1 or 2 random regions from 40 normal-hearing listeners. While the recognition scores ranged from 8.2% to 16.3% when E information from only one region was available, the scores ranged from 57.9% to 87.7% when E information from two frequency regions was presented, suggesting a synergistic effect among the temporal E cues in different frequency regions. Next, the relative contributions of the E information from the five frequency regions to sentence perception were computed using a least-squares approach. The results demonstrated that, for Mandarin Chinese, a tonal language, the temporal E cues of Frequency Region 1 (80–502 Hz) and Region 3 (1,022–1,913 Hz) contributed more to the intelligence of sentence recognition than other regions, particularly the region of 80–502 Hz, which contained fundamental frequency (F0) information.